Introduction

To assure the fluorophores were functional, the emission spectra were recorded at the given excitation wavelength.

Methods

Parts encoding the four different fluorophores, including promoter, RBS and terminators, were expressed in E. coli BL21 strain. GFP, mVenus, mKate and mCerulean were all expressed under the same strong constitutive promoter, J23100. Additionally, parts were constructed with GFP under control of promoters with different strengths, in order to investigate the influence of different fluorophore concentrations.

All of the mentioned biobricks were expressed in E. coli BL21. After growing in LB medium to an OD600 of ~1, the cells were washed and resuspended in the same volume of PBS. Aliquots of 100 µL were put in a 96 well plate.

The emission spectrum of each fluorophore was determined by exciting at a given wavelength and measuring the output intensity at a range of wavelengths. Because of this, the emission at a wavelength too close to the excitation could not be measured. This can be seen in the figures, where the left half of the emission peaks could not be measured. Especially for mVenus, where the excitation and emission wavelengths are very close together.

Results and Discussion

Fluorophores expressed under strong constitutive promoter

As all fluorophores were expressed under the strong constitutive promoter J23100, they were expected to show a strong fluorescence without the need of induction. Figure 1 shows that this was the case for GFP, mVenus and mCerulean. mKate, however, did not show any fluorescent activity and was therefore not used in the subsequent steps of the project.

GFP expressed under constitutive promoters of different strengths.

Not only did we express GFP under the strong promoter J23100, but also under less strong promoters J23113, J23117, J23105, and J23108.

For comparison, the emission was normalized by dividing by OD. All strains were measured in the same dilution, in order to make the results reproducible. The emission intensity is as expected, corresponding to the strength of the promoters. All fluorophore spectra were also recorded in a dilution more suited for their emission intensity and normalized by 1 to the maximum intensity of each construct (Figure 3). From this figure we can conclude that all GFP biobricks function properly.

Back to Top

Introduction

As consitutive expression can sometimes be hard on the cells, we investigated the effect of the different consitutive promoters on cell growth. In a 24 hour kinetic cycle alterating shaking at 37°C with fluorescence and optical density measurements, we investigated whether this was the case.

Methods

An overnight culture in eM9 medium was inoculated in fresh eM9 to an OD600 of 0.1 in a 96 well plate. The emission at 522 nm was measured every 15 minutes. Measurements were done in quadruplicate with pure eM9 as a blank.

Results and Discussion

Figure 1 shows the 24 hour measurement of optical density and fluorescence intensity. The final OD600 is approximately equal for all different strains, suggesting that the level of constitutive expression was not influencing the growth. Furthermore, fluorescence intensity drops after the exponential growth phase, suggesting that GFP is being broken down by proteases as a response to nutrient limitation. After this event, growth continues at a slower pace, while GFP activity keeps decreasing. All in all, constitutive expression of GFP does not seem to have a detrimental effect on cell growth during exponential phase.

Cells expressing mCerulean or mVenus, however, seem to be having a longer lag phase before exponential growth starts (Figure 2). Also, they reach a lower final OD600 than the ones expressing GFP. These fluorophores might be slightly harmfull for cell growth. Nonetheless, they do grow and exhibit fluorescence, so they can be used in further experiments.

Back to Top

Introduction

After transforming E. coli with the different silicatein BioBricks, we wanted to confirm and image whether the cell had indeed synthesized a layer of polysilicate around itself. Before using any advanced imaging techniques, we first used a much simpler technique. It is possible to stain polysilicate depositions with the fluorescent stain Rhodamine 123. This stain has shown to bind specifically to polysilicate (Li, Chu, & Lee, 1989). Since the stain is fluorescent, we can image it with a simple fluorescence microscope. If a cell has a polysilicate layer, the Rhodamine 123 will bind to it which we can image.

Methods

The experiment was performed using E. coli BL21 with the following plasmids and conditions. All genes are under an inducible promoter (Lac-promoter), to do so it was cloned in a backbone containing the promoter and all machinery necessary for it to work. This backbone was obtained from pBbS5a-RFP, a gift from Jay Keasling (Addgene plasmid # 35283) (Lee et al., 2011).

Plasmid(s)	IPTG	Silicic acid	Rhodamine 123
OmpA-Silicatein	+	+	+
Silicatein	+	+	+
INP-Silicatein	+	+	+
OmpA-Silicatein (negative control)	+	-	+
OmpA-Silicatein	+	+	-

The cells were stained with 0.1 vol% Rhodamine 123. And washed 5 times with PBS, prior to imaging in a fluorescence microscope at 488nm (Li et al., 1989; Müller et al., 2005). Both widefield- (light microscopy) and fluorescence microscopy were used to image the cells.

Results and discussion

The stained cells were first imaged at maximum excitation intensity. At this excitation energy, only OmpA-silicatein showed fluorescence specifically localized at the cells and not in the medium. Silicatein, INP-silicatein and the negative control (OmpA-silicatein without silicic acid) all caused overexposure of the camera. In figure 1 the imaging results of OmpA and the negative control (OmpA-silicatein without silicic acid) are shown. The cells that were not stained with Rhodamine 123 showed no measurable fluorescence. (data not shown).

Since Silicatein, INP-silicatein and the negative control all caused overexposure of the camera, they all had the same output. We can thus not draw any conclusions for these strains. Therefore, samples where overexposure was observed were imaged again at only 1/3 of the excitation energy. The imaging results are displayed in figure 2.

From figures 1 and 2 we can see that the strain transformed with OmpA-silicatein clearly has a different output from the negative control (figure 1). The fluorescence is only localized at the cells. This might mean that the Rhodamine 123 has stained these cells and therefore the OmpA-silicatein cells could have the polysilicate layer around their membranes. We cannot distinguish a clear difference between silicatein, INP-silicatein and the negative control (figure 2). The entire medium is fluorescent, which causes overexposure of the camera at high excitation intensity. This might mean that the Rhodamine 123 is not specifically located at the cell walls, but still dissolved in the medium.

Because the cells were analyzed at different excitation energies, so different camera settings, it is impossible to draw a conclusion from solely the images. Therefore, we analyzed the pictures using ImageJ. We measured the intensity of the fluorescence of the cells and the intensity of the fluorescence from the background. Then we normalized the intensity of the cells over the intensity of the background (in this case, the medium). This way, the intensity of the background of all samples becomes 1 and the intensity of the fluorescent cells has a relative value to the background signal. With this method, we are able to compare the fluorescence intensities of the cells, despite the different camera settings. The results of this calculation are shown in figure 3.

In figure 3 we see that the fluorescence intensity of the OmpA-silicatein strain is significantly higher than for the the silicatein and INP-silicatein strains. The silicatein and INP-silicatein strains have a fluorescence intensity comparable to the intensity of the medium (the background intensity). The negative controls (no silicic acid added, measured at two intensities) have the same output as silicatein and INP-silicatein. Therefore, we can conclude that these strains are not able to synthesize a polysilicate layer. However, we can conclude that the E. coli strain transformed with the OmpA-silicatein plasmid is successfully able to synthesize a layer of polysilicate around its membrane.

Back to Top

Introduction

Electron microscopy is an imaging technique that can acquire images with resolutions far above the optical limit due to the small wavelength of electrons (Nellist, 2011). An electron gun forms an electron beam which is focused and accelerated by magnetic lenses. When the electrons reach the sample, there are several interactions possible with the sample (figure 1). These various interactions can be used for different techniques. We have used high-angle annular dark-filed transmission electron microscopy (HAAFD-TEM) to make an image of the sample, energy dispersive x-ray spectroscopy to analyze the element composition of the sample and scanning electron microscopy (SEM) to analyze the phenotype of the samples.

In HAAFD-TEM electrons pass through the sample and are inelastically scattered at a high angle and acquired by a detector. The amount of electrons scattered at a high angle depends on the thickness and the material of the sample. When the thickness is high, the sample scatters a lot it will appear bright in the image.

Energy dispersive x-ray spectroscopy (EDX) is an analytical technique to determine the elements present in the sample (Friel et al., 2006). When the electron beam interacts with the sample x-rays can emerge from the sample. These electrons can be detected and are characteristic for the element which emitted the x-ray. In that way we can specifically determine which elements are present in our sample.

Methods

The experiment were performed using E. coli BL21 cells with the plasmid containing OmpA-Silicatein. Two samples were made where the first sample was induced with IPTG but no silicic acid was added and a second sample which was both induced with IPTG and silicic acid was added, therefore this sample will have a polysilicate layer. The samples were prepared by fixation using 1% polylysine on the surface of a quantifoil carbon grid.

Results and discussion

Both samples with and without silicic acid added were imaged using HAAFD-TEM and energy dispersive x-ray spectroscopy (figure 2). The cells are fixed at a Quantifoil carbon grid. In figure 2A and 2C the white structure is a cell laying on a hole in the grid. In each sample, we measured elemental composition of our sample including the silicon content (figure 2 B,D) the blue spots in these images indicate where silicon is detected. The grid itself already contains silicon but in the holes of the grid no silicon is present (figure 2 B,D). Therefore we measured the pressence of silicon in bacteria laying on a hole on in the grid to make sure we do not have background silicon signal from the grid.

For the sample where no silicic acid is added (figure 2 A,B), we can see some silicon present at the position of the cell. However, there is a significant increase in silicon detected for the sample where silicic acid was added to the sample (figure 2D). This shows that silicon co-localizes with the cell which means there is indeed a polysilica layer formed by the bacteria.

Back to Top

Introduction

The Scanning Electron Microscope, or SEM as it is called, is a widely used imaging technique in science. In SEM, a beam of electrons is accelerated by high voltage and it is directed towards the sample with speeds up to \( 1.64 \times 10^5 km\). When the beam hits the sample, x-rays, backscattered electrons and secondary electrons are generated. For imaging, SEM uses the secondary electrons. Those secondary electrons are then collected and their kinetic energy is transformed to an image. (Kwakernaak & Sloof, 2005) It is highly used due to its very high spatial resolution and Depth of Focus (DoF). SEM can reach resolution of 5 nm and DoF about 0.5 mm, compared to an optical microscope that can reach a spatial resolution of \( 0.2 \mu m \) and DoF \( 1 \mu m \). With SEM it is only possible to image solid samples because the measurements take place under vacuum.

With SEM, we hope to see the impact of the polysilicate layer on the phenotype of the cell. If we want to make microlenses, the surface of the cell should be smooth, despite the silicatein layer. Any deformations in the layer can cause optical aberrations. Since SEM has a very high spatial resolution, we can use this technique to image what our silicatein-covered cells look like and to see what effect the polysilicate layer has on the strains.

Methods

The polysilicate layer around the cells was prepared according to the polysilicate layer protocol. As a negative control, uninduces cultures (no IPTG) and cultures without substrate (no sodium silicate) were used. The samples were fixed with gluteraldehyde and imaged with SEM. We used an FEI Niva Nano 450 SEM, under high vacuum.

Results and Discussion

Figure 1 shows SEM images taken of samples in the presence (A, B) or absence (C, D) of silic acid.

First of all, since we know from a previous rhodamine 123 staining experiments that cells with OmpA-silicatein have a polysilicate layer, therefore it does not seem to influence the cell shape. However, the cells that are expected to have a polysilicate layer appear to be somewhat fused together. According to Müller et al (2008), cells possessing a polysilicate layer appear to be fused by a viscous cover. This can, however, also be the result of limited imaging resolution. When using titanium oxide as a substrate, Curnow et al (2005) reported large aggregates visible by SEM. This was not observed in the current experiment. Therefore, we can conclude that the polysilicate layer does not form aggregated or influence the shape of the cell, but form a homogeneous layer around the cell.

Back to Top

Introduction

The physical properties of polysilica are different then those of the cell. We wanted to determine how the physical properties of the cell changes when encapsulated by a layer of polysilica. Therefore we determine the stiffness of cells with and without a layer of polysilica using atomic force microscopy (AFM). The stiffness is the rigidity of the a material, describing how easy the cell is deformed.

AFM is a technique used to determine surface characteristics of a sample. In AFM a very sharp tip scans over the surface of the sample. While scanning over the surface every height change is detected by a laser which is reflected by the cantilever on a position-sensitive photodetector (Figure 1). We used the PeakForce QNM mode of AFM, in which also surface characteristics like the stiffness and deformation can be determined. The stiffness is computed from the relation between the force at which the cantilever pushes on the sample, the adhesion force and the deformation of the material.

Methods

Experiments were performed using E. coli BL21 cells transformed with the plasmid containing the OmpA-Silicatein gene and induced with IPTG. Silicic acid was added to one sample to form the polysilicate layer around the cell. Another sample with no silicic acid added was used as a control. The cells weres spun down and resuspended in MilliQ and fixated on a glass slide using 1% ABTES.

Results and discussion

We have imaged two samples with AFM. The first sample is OmpA fused with silicatein with silic acid added so that the cell can encapsulate itself with biosilica (figure 3A-B). The second sample, that wa used as a control, did not contain silicic acid (figure 3C-D). From both samples a height map (figure 3A,C) and the stiffness (figure 3B,D) was determined. Both samples were fixed on a glass slide in the same way. Due to a tip change is there a factor 10 difference between the stiffness of the glass slide of both samples.

Multiple cells (n=3) were imaged and the relative stiffness of the cell compared to the stiffness of the glass slide was determined (figure 3E). We found that cells covered with a layer of polysilicate have a stiffness of 0.43 compared to the glass slide and the cells without a layer of polysilicate have a stiffness of 0.16 compared to the glass slide. We can see that due to the encapsulation of the cells in polysilica the stiffness of the cells increases significantly.

Back to Top

Introduction

Since the silicatein expressing cells are to cover themselves in polysilicate, their nutrient supply might be limited by diffusion, which can eventually result in cell death. To investigate whether this is indeed the case,a growth study was performed.

Methods

Cells containing the three different silicatein biobricks were grown overnight in selective LB. They were transfered to fresh medium and grown until in exponential phase. Then IPTG was added to induce expression. After a subsequent incubation of three hours, the medium was supplemented with silicic acid as substrate for silicatein. During the following five hours samples were taken, of which a 10^-6 dilution was plated on selective LB plates. Colony forming units (cfu) were counted the day after.

Results and Discussion

Cells expressing either silicatein (Sil), silicatein fused to INP (INP Sil) or silicatein fused to OmpA (OmpA Sil) were tested. As a negative control, OmpA Sil Taur expressing cells without silicic acid were used. After one hour no colonies were observed on the plates on which the cultures with silicic acid were plated (Figure 1). The cultures without silicic acid continued to grow until after five hours.

This figure suggests that either the polysilicate layer inhibits nutrient diffusion into the cell, unlike reported by Müller et al. (2008), or the sodium silicate has a detrimental effect on growth.

Back to Top

Introduction

Building an optical setup is a very precise work. First, it is important to calculate the positions of all components in such a way that the laser beam will reach your sample, and the light emitted by the sample consequently reaches the detectors of the camera and spectrometer. Once this is done, all optical components are positioned on an optical table. This special table is made to prevent vibrations in you system and has mounting holes so all optical components can be screwed into place. Once the components are positioned on the table, the alignment begins. In this step, the beam coming from the laser is guided throughout the system. By slightly adjusting and repositioning all optical components the light is guided through the components until it reaches the detector of the camera. It is essential that the components are aligned correctly and are free of vibrations, because this could change the path of the light.

After tens of hours of carefully placing components and aligning the light through the setup, we managed to direct the light from the laser, through all components onto the detector of the camera. However, this does not necessarily mean that when we add fluorescent cells to the setup it we are able to image the cells with the setup. Any error in the setup could cause it not to work. Therefore, we first had to confirm whether our setup was working.

Methods

In order to confirm whether the setup was working, we used E. coli BL21 cells that were transformed with our constitutive mCerulean BioBrick. In previous fluorescent plate reader experiments, we have confirmed that these cells are able to fluoresce and that they can be excited at 405nm, the wavelength of our laser. To make sure the only output we were measuring was fluorescence, and not any ‘leakage’ of light from our laser beam, we also tested cells that were not transformed with the mCerulean BioBrick. These cells are not able to fluoresce after excitation at 405nm, so if the setup is working properly we should not get a signal from these cells.

The cells were adhered to the microscope slides using 3% agarose pads and imaged at an excitation intensity of 0.5 mW. This energy is low enough to not instantly photobleach the proteins, but observe fluorescence clearly. Focusing on the cells was done manually with a 50x oil-immersed objective. The cells were excited with a Coherent OBIS LX 405nm laser and images were taken using a DeltaPix Invenio III CCD camera.

Results and discussion

Imaging the cells with the setup yielded the following results:

As we can see in figure 2B, the cells transformed with a gene for the fluorescent protein mCerulean are clearly visible. When using a strain transformed with a plasmid that is not known to cause the cells to fluoresce, in this case OmpA-silicatein, no light was observed, as shown in figure 2A. To confirm that this sample did contain cells, we also inspected the sample with light microscopy, as shown in figure 3.

In figure 3, we see that the negative control did contain cells, but we could not observe a fluorescent signal. From this we can conclude that we successfully built a setup that is able to observe and measure fluorescence in a cell. There is no leakage of light of our excitation laser in the camera, since we do not observe anything when we use non-fluorescent cells. Also at a higher excitation energy (50 mW) we did not observe anything on the camera. Therefore we can conclude that our setup works as expected as it is indeed able to measure fluorescence without measuring other light sources.

For this experiment we used E. coli transformed with OmpA-silicatein that was induced and incubated in silicic acid, so it would contain the silica layer. We deliberately used this strain to test whether the silica or the cells had any autofluorescence that could interfere with our laser experiments. We did not observe any fluorescent signal for these cells, so we can conclude that the silica layer does not have any autofluorescence at 405 nm. Therefore, these cells are suitable for the laser experiments.

Back to Top

Introduction

One of the biggest differences between a laser and fluorescence is the amount of emitted photons. We can measure the amount of emitted photons by measuring the intensity of emitted light. The intensity of fluorescence increases linearly with the excitation energy. However, at a certain excitation energy it will reach the so-called ‘laser threshold’ and stimulated emission, and thus lasing will occur. From this point on, the output intensity of the fluorophores will still increase linearly, but with a much steeper slope, as shown in figure 1.

We investigated whether we could find the same relation between input and output intensity for our Biolaser cells, using our self-built setup, to investigate whether our cells were able to emit a kind of laser light.

Methods

For this experiment we used E. coli BL21 transformed with the following BioBricks:

Plasmid(s) and conditions	Function
mCerulean (constitutively expressed)	Fluorescence
mCerulean (constitutive) + OmpA-silicatein (induced, incubated in silicic acid)	Biolaser
OmpA-silicatein (induced, incubated in silicic acid)	Negative control (no fluorescence)

The cells were adhered on a microscope slide using 3% agarose pads. The cells were excited at a wavelength of 405 nm. The cells were imaged at excitation energies of 0.1 mW, 0.5 mW, 0.7 mW, 1 mW, 2 mW, 5 mW and 10 mW. These images were analysed using ImageJ to determine the output intensity and corrected for background noise (McCloy et al., 2014).

Results and discussion

The output intensities of our cells were plotted against the excitation power to determine whether our cells emitted laser-like light. The results are shown in figure 2.

In figure 2, the measured intensity is plotted against the excitation power (black dots). Since we expect a linear relation, the expected increase of intensity of fluorescence (blue line) and the expected increase of intensity for lasing (red line) was also plotted. The expected fluorescence is fitted to the fluorescence of the mCerulean, the expected lasing pattern is estimated from this. We show two duplicate experiments per strain. Comparing the measurement data to the expected relations, we can not see a convincing result that our cells are lasing. Above an excitation power of 2 mW we see that in all cases the fluorescence stays at the same level. This means that the intensity is not increasing anymore and we believe that we observe photobleaching. So, both the fluorescent cells as well as the laser cells do not emit any laser-like light. For the fluorescent cells that were transformed with solely mCerulean, this result is as expected, these cells do not have any extra modifications that should cause them to emit laser light. The strain that was transformed with both mCerulean and OmpA-silicatein had a glass shell around the cell that could cause the cells to emit laser light. This effect was not observed. The strain transformed with solely OmpA-silicatein did not emit any light, as expected.

From this experiment we can not see any clear proof of lasing in our cells. This could be both due to our setup or due to the constructed cells. Previous studies show that lasing susually occurs at an excitation power under 1mW (Gather & Yun, 2011; Fan & Yun, 2014). However, up to this excitation power the Biolaser cells follow the same slope as the fluorescent cells, which means that no lasing occurred. From 2 mW and higher the cells most likely bleach. Since we expected the cells to lase at an excitation energy under 1 mW, we did not take such high energies into account while designing our setup. Even though we exposed the cells to the excitation laser for a very short time, the power was probably so high that the cells eventually photobleached anyway. It is also possible the maximum fluorescence intensity of the cells was reached, and therefore the intensity doesn't increase.

Even though the excitation laser eventually most likely bleached or saturated the fluorophores, this was probably not the reason why lasing didn’t work. Lasing should have occurred at an excitation energy between 0 and 1 mW, and at these energies the fluorophores did not bleach. It is possible that our cells were lasing at the low energy prior to bleaching or saturation, but the system was not sensitive enough to measure this. However, we did find a more probable reason why the cells did not lase. Modeling showed, that in a cavity as big as E. coli (around 1 µm) we need an intracellular fluorophore concentration of 0.1 M to get lasing. In our cells the maximum concentration we can achieve is in the nM to µM range and therefore lasing physically not possible in our cells. To get lasing, we would either need much larger cells or a much higher concentration of fluorophores. Therefore, it was most likely not due to the self-built setup nor our synthetic biology design of a biolaser that we did not observe lasing.

Back to Top

Introduction

When making biological lenses, the shape of the lens is of crucial importance. E. coli is a rod-shaped organism, so it’s not symmetrical along all axes. Shining light on the round parts of E. coli has a different effect on the focusing of light than shining light on the long sides, see figure 1. More information on this can be found on the modeling page.

When we want to use our microlenses in more advanced optical systems, such as microscopes or cameras, we need to make sure that the variation between the different lenses is minimized. Manufacturers of optical systems do not accept a high aberration between different lenses, so it’s crucial for us to be able to control the shape of our lenses. We have decided to engineer E. coli in such a way that it becomes spherical. This way we are able to create spherical lenses. Apart from the fact that it is crucial to be able to control cell shape, round cells offer the advantage of being symmetrical along all axes, so the orientation of your lens does not matter for the optical properties.

In order to create spherical E. coli, we overexpress the BolA gene. BolA is a gene that controls the morphology of E. coli in the stress response (Santos, Freire, Vicente, & Arraiano, 1999). By overexpressing this gene at 37°C, the rod-shaped E. coli cells, will become round (Aldea, Hernandez-Chico, De La Campa, Kushner, & Vicente, 1988). We will express this gene both under a constitutive promoter (Part K1890031), as well as an inducible promoter (Part K1890030). When we express both the BolA gene as well as silicatein, we are able to construct round cells, coated in glass.

Methods

The expected phenotype of the cells expressing BolA is very different from the phenotype of wildtype E. coli. If the gene is successfully overexpressed, the cells become round, which we can easily observe under a widefield microscope. In widefield microscopy, the whole sample is simultaneously illuminated using a white light source so the phenotype of the sample can be inspected. This is comparable to normal light microscopy.

In order to obtain round cells we tested transforming E. coli BL21 with BolA under both a constitutive promoter (J23100) and an inducible promoter (Lac-promoter), to do so it was cloned in a backbone containing the promoter and all machinery necessary for it to work. This backbone was obtained from pBbA5c-RFP, a gift from Jay Keasling (Addgene plasmid # 35281) (Lee et al., 2011). Furthermore, we tried if transforming a strain with both BolA and the OmpA-silicatein fusion plasmid yielded round, glass covered cells. The following strains and conditions were tested under the widefield microscope:

Plasmid(s)	IPTG	Silicic acid
OmpA-silicatein (inducible)	+	+
Lac-BolA (inducible)	+	-
J23100-BolA (constitutive)	-	-
OmpA-Silicatein (T. aurantia) + Lac-BolA (inducible)	+	+

The cells were heat-fixed on a slide and observed under the widefield microscope at 37°C. Furthermore, imaged E. coli with and without the BolA-gene (induced) under SEM. The samples were fixed with gluteraldehyde and imaged with SEM. We used an FEI Niva Nano 450 SEM, under high vacuum.

Results and discussion

The four different strains were imaged under the widefield microscope, the taken images are shown in figure 2.

In figure 2, the widefield images of the four tested strains are shown. In figure 2A we can see cells with the characteristic wildtype E. coli phenotype; the cells are rod-shaped. Figure 2B shows that induction of the cells transformed wil BolA under the inducible Lac-promoter indeed has changed the phenotype of the cell. The cells have clearly become spherical. Constitutive expression of the BolA gene, as shown in 1C has the opposite result: the cells are rod-shaped and elongated. This is probably because of the stress the plasmid puts on the cells. As mentioned before, BolA is a gene involved in the stress response of E. coli that changes the morphology of the cells. A too high expression of the gene could therefore change the morphology in an unexpected way, such as elongation, a phenomena that is often observed in E. coli (Höltje, 1998). Since we require sphere-shaped cells, constitutive expression of the BolA-gene is not desired. Co-expression of the OmpA-silicatein fusion plasmid and the inducible BolA plasmid also yielded round cells, as seen in figure 2D.

To get a more detailed view of the shape of the cells transformed with BolA, we also imaged the cells using SEM, the results are shown in figure 3.

In figure 3, we see that the cell shape of the cells transformed with BolA is indeed more spherical than the wildtype phenotype of E. coli. Also, we see that the size of the cells increased. Using ImageJ we analyzed the average size of 10 cells:

In the table we see the average diameter of the cells increased. The average size of microlenses is in the order of tens or hundreds of micrometers, because with current techniques it's difficult and costly to produce smaller microlenses (Krupenkin, Yang, & Mach, 2003). The average sizes of photodetectors are in the nano- to micrometer range (Yang, Shtein, & Forrest, 2005). Our spherical microlenses approach this size, and are therefore usefulm since you could place one microlens over one photovoltaic cell, increasing the efficiency of the photovoltaic cell.

So, by inducing the expression of BolA, we are indeed able to control the shape of E. coli and turn the cell into a sphere. Also, transforming a cell with both silicatein-OmpA and BolA yields round cells, so the formation of the glass layer does not distort the cell shape. We have also shown that even though cell size increases upon expression of BolA, the size of our biolenses is smaller than the size of conventional microlenses. People have not been able to produce such small lenses in a cost-effective way, so our microlenses could be a possible solution to this.

Being able to control cell shape is of major importance if we want to create biological lenses, since the lenses are desired in various sizes and shapes. Especially spherical lenses are useful since they are symmetrical and therefore do not require a specific orientation; they will focus the light in the same way whatever their orientation is. From figure 2 we can see that the cells are not perfectly homogeneously shaped, there is some variation between the shape of the different cells. This variation is even clearer for the cells that contain the silica layer. This is possibly because two plasmids with a lac promoter put a great strain on the cells, resulting in a greater variation. For precision optics, it is extremely important that there is little to none variation between the lenses. Therefore, it’s recommended to do more research in controlling cell shape. However, since there is always a variation in gene expression between cells it will be wiser to conduct research into selecting or sorting the cells on their size or shape. A possible solution could be sorting the cells using FACS. As a future experiment, we could stain the silica-coated cells with Rhodamine 123 and let a FACS-machine sort the cells on their phenotype.

Back to Top

Introduction

We wanted to test our biolenses in a real-world application: solar cells. This was chosen, since it was shown to make solar cells more efficient while using microlenses (Jutteau et al., 2015). Solar panels are currently unable to use all available solar energy, since parts of the light is lost due to non-focused beams. To decrease the losses, light needs to be focused, which can be done by lenses. Jutteau had shown already that solar panels are increased in efficiency while using microlenses. However, these microlenses are chemically produced, meaning under harsh environmental conditions. A high temperature combined with low pressures and harmful chemicals to produce these microlenses do not contribute positively in the total image of “environmentally friendly”.

In our opinion, the eventual product – microlenses – was promising. However, the production method did not add up with the purpose of increasing the potential of solar energy and thereby making general energy use more environmentally friendly. Our biologically produced E. coli cells covered in silicate were produced under physiological mild conditions (37 ºC, 1 atm and no harsh chemicals) and would fit much better in the image of environmental friendliness.

To put our cells to the test and see if these are just as promising in increasing the efficiency of solar cells as chemically produced microlenses, we have called for help in the department Photovoltaic Materials and Devices at the TU Delft. Here, prof. Dr. Arno Smets, had introduced us to Stefaan Heirman, a technician at the same research group. Stefaan helped us in our solar cell experiments.

Since Stefaan, including the to be used solar simulator and all other necessary equipment were at another faculty of the TU Delft, our cells had to leave the ML-1 lab. Here, safety regulations from the executed practices were incorporated in the labwork, since the biolenses should be outside the lab – into the environment as well as soon as they could be used on solar panels. Even though the growth study of our in silicate covered E. coli cells did not show any growth anymore after 1 hour, due to safety regulations we have decided to used an additional method to sterilize our biolenses.

To sterilize, different options are available. For example alcohols, chlorine and chlorine compounds are used within healthcare and autoclaving is a much-used method in laboratory settings. Besides that, UV sterilization is a much-used method in water treatment (Chang et al., 1985; Hijnen et al., 2006; O’Flaherty et al., 2016). This is used to remove among others E. coli. Since we do not want any chemicals to potentially interfere with our biolenses, we have chosen not to use any of these chemicals and to test whether autoclaving or UV sterilization would be a good option.

Methods

Autoclaving biolenses

Aliquots of E. colicells transformed with OmpA-silicatein were induced with silicic acid and subsequently autoclaved for 20 minutes at 121 ºC before imaged with scanning electron microscopy (SEM).

UV sterilization

For this experiment, E. coli cells transformed with BolA pBbA5c were used, as the results of the growth experiments just showed that cells covered in silicate were not viable. Since we did want a positive (absolutely living) control, we have chosen the BolA strain. For

A volume of 30 µL E. coli BL21 cells that were transformed with BolA in pBbA5c was plated on LB plates complemented with chloramphenicol. As controls for sterilization aliquots of 500 µL of the same cells were either heated at 95 °C for 10 minutes or washed with 500 µL 70% ethanol for three times. For UV sterilization, light bulbs emitting 254 nm were used and samples were exposed to it for 3 minutes. After sterilization, also 30 µL of these cells were plated on LB+Cm plates. Besides that, 30 µL of non-sterilized cells were plated as well. Of the last mentioned cells, it was expected for them to be alive and grow after overnight incubation.

Two of the plates containing BolA transformed cells were UV sterilized in a Stratagene UV Stratalinker 1800. The manufacturer recommended power settings of 120000 µJ. In our experiment we have used a power of 120000 µJ and 240000µJ to sterilize our cells.

All plates were incubated overnight at 37 °C.

Results

Autoclaving

Cells containing the OmpA-silicatein gene were autoclaved and subsequently imaged with SEM.

Since the SEM showed the cells were collapsed, and in no way could function as biolenses, another sterilization method was used.

UV sterilization

To test the effectivity of UV sterilization on our E. coli cells, a growth study after UV sterilization was performed. For this 30 µL BL21 cells transformed with BolA in pBbA5c was plated on LB plates complemented with chloramphenicol. These plates were UV sterilized in a Strategene UV Stratalinker 1800. As positive controls the same cells were heated for 10 minutes at 95 ºC or washed three times with 500 µL ethanol. As negative control the same cells were plated on a selective plate.

None of the sterilized plates showed any growth, while the non sterilized plate did yield colonies.

Now it was shown that the E. coli cells are not viable after UV sterilization, we also wanted to see the influence of UV sterilization on the physical properties of the silica layer. Since the silica layer already ensured unviability of the cells, UV sterilization could be seen as an extra safety precaution to compline to current synthetic biology safety regulations. For this, we have UV sterilized E coli cells transformed with OmpA-silicatein, using 120000 µJ for 3 minutes and these were subsequently imaged with SEM (figure 2).

As shown in figure 2, the with silicate capturing of the cells remain undamaged after UV sterilization and could be used as biolenses now.

Discussion

In this experiment, different methods to sterilize our biolenses were tested, as we wanted to be able to put our biological microlenses in real-world conditions. For this, we have both autoclaved and UV sterilized our cells. From the growth study, it can be concluded that the biolenses were successfully sterilized by UV sterilization. Besides that, the shape of the biolenses was not affected by UV sterilization as opposed to autoclaving.

Back to Top

Introduction

One of the most common and useful methods of characterization of optical materials is done by spectroscopy. Using spectroscopy we can measure how much light can be transmitted through materials. The device used for those measurements is called spectrometer because it scans the electromagnetic spectrum and detects how much of the initial radiation passes through the materials, which are in our case our samples. For those measurements samples are loaded on glass plates and dried overnight, forming a solid material, then a beam of light is shined through it. The beam enters an integrating sphere in the back of the sample that doesn’t let any light to escape and guides it to the photodetector that measures how much light is still present after the sample. If we want to use our polysilicate-covered cells in an optical device or on a solar cell it is important to know whether the bacteria absorb light. If they do, they will significantly decrease the efficiency of the solar cells or optical device. Since our goal is to capture more light, we need to make sure that the cells do not block any light.

Methods

In order to do the spectroscopy measurements of our cells, the following samples were prepared:

Plasmid(s)	IPTG	Silicic acid
OmpA-Silicatein	+	-
OmpA-Silicatein	-	-
BolA + OmpA-Silicatein	+	-
BolA + OmpA-Silicatein	+	+
OmpA-Silicatein	+	+

The bacteria were killed using UV sterilization and placed on glass pieces to dry overnight. Since our biolenses were placed on a glass plate instead of directly applied to the solar cell, another parameter was added: the glass plate could potentially also influence the spectroscopy measurements. Taking this into account, a control measurement of a clean glass plate had to be performed and then subtracted from the measurements of our biological samples. This way we could measure the effect of our lenses on the light detected by the solar cells, which we call the ‘spectral response’. The spectral response was scanned from 300 nm to 1200 nm, since this is the range of wavelengths measured in the solar cells. Because we are not so interested about the exact spectral response in each wavelength we averaged the transmission data.

Results & discussion

The amount of light that was absorbed by our cells was measured, as well as the absorbance of the substrate - a glass slide - on which our cells were placed and dried. The absorbance of the substrate was subtracted from the measurements done with our cells to get the absorbance of solely our biolenses. The amount of light that is not transmitted by our cells is presented in figure 1 below.

Figure 1 demonstrates how much light is not transmitted by our cells. The light that is not transmitted is either absorbed or reflected. Nevertheless, from figure 1 we can see that the light that was shined on the cells and did not pass through was less than one percent. To put things in perspective, figure 2 demonstrates the light that gets transmitted through our devices.

In figure 2 we see that more than 99% of the light is transmitted through our cells. We do not see a significant difference between polysilicate-coated cells and uncoated cells. The intention of the polysilicate layer was to focus the light, not to increase transmittance through the cells. However, due to the extra layer it was hypothesized that our biolenses would absorb more light compared to uncoated cells. The fact that we see no significant difference means that the polysilicate layer does not have a detrimental effect on the optical properties of the cells.

Concluding, the spectroscopy experiments showed that our biolenses almost have no influence on the amount of light that was passed through. More than 99% of the initial light is transmitted through our microlenses, making them a good option for any kind of optical application. Moreover, the results obtained from this experiments verify the assumption made in the modeling of microlenses that we have non-absorbing materials, so the imaginary part of the refractive index was set to zero. Furthermore, it verifies the scattering model where we concluded that we have hardly any backscattering from our devices.

Back to Top

Introduction

Solar panels are used to convert the potential of solar energy into electrical energy. A lot of innovation had already happened in this field, however the efficiencies can still use quite some improvement. As it was stated by Jutteau et al. (2015), the efficiency of solar cells could be increased significantly by adding a layer of microlenses. As parts of the available light get lost due to the fact that solar light is parallel and non-focused, the amount of electrical energy that can be retrieved from the sun could be higher. This limitation can be solved by applying a layer of microlenses onto the solar cell that focus the light in the solar panel, as shown by Jutteau et al. (2015).

The microlenses used in the research of Jutteau et al. (2015) are chemically produced under harsh conditions, such as extremely high temperatures, low pressure and caustic chemicals. This does, in our opinion, not complement the aim of being more environmentally friendly.

Our E. coli cells covered in polysilicate however, could suppose a biological alternative for these chemically produced microlenses. The biological microlenses, were produced under physiological mild conditions (37 ºC, 1 atm and no harsh chemicals) and would fit much better in the image of environmental friendliness.

To put our cells to test and see if these are just as promising in increasing the efficiency of solar cells as chemically produced microlenses, we have called for help in the department Photovoltaic Materials and Devices at the TU Delft. Here, prof. Dr. Arno Smets (see acknowledgements), introduced us to Stefaan Heirman, a technician at the same research group who helped us in our solar cell experiments. After sterilization, we compared efficiencies of solar cells using a solar simulator with and without our biologically produced microlenses.

Methods

There are different ways to measure the output of solar cells with accuracy and reproducibility. The most used characteristic of PVs is their efficiency (Power Conversion Efficiency or PCE). It is defined as the energy produced by the device with a set amount of light from an external light source. There is a standardized protocol for this type of measurement, enabling researchers to simulate ‘real-world’ conditions and compare different solar cells in different setups.

In order to achieve these reproducible and accurate ‘real world’ measurements of efficiency, the previously mentioned protocols need to be used, as well as a certain type of equipment. The equipment used to simulate the aforementioned conditions is called a solar simulator, which is basically a big Xenon lamp with well-defined light power output. In order to calibrate the simulator a ‘standard’ solar cell is used with well-defined properties to measure the exact output of the simulator in each measurement. Finally, the tested device must be placed perpendicular to the light with nothing restricting the light path and kept still during the whole testing. Using this set of equipment, we are sure that we conduct all the experiments under the same scientifically accepted conditions so we remove the uncertainty factor from our experimental results.

We tested the following cells in this ‘real-world’ solar setup:

Results and discussion

In order to do the spectroscopy measurements of our cells, the following samples were prepared:

Plasmid(s)	IPTG	Silicic acid
OmpA-Silicatein	+	-
OmpA-Silicatein	-	-
BolA + OmpA-Silicatein	+	-
BolA + OmpA-Silicatein	+	+
OmpA-Silicatein	+	+

The bacteria were killed using UV sterilization and placed on glass slides to dry overnight. The Power conversion efficiency of the uncoated solar cell was measured, as well as the efficiency of the solar cell with an empty glass slide (negative control) and the efficiency of the solar cell with a glass slide coated with bacteria.

Results and discussion

The measured Power conversion efficiencies of the solar cell and the solar cell with various coatings is shown in figure 2.

In figure 2, we see that the uncoated solar cell has a power conversion efficiency of around 15%. Since our cells were affixed on a glass slide, we first had to measure the impact of placing a glass slide on the solar cell, so we can distinguish the effect that our cells have on the efficiency from the effect that the glass slide might have. We see that placing a glass slide, the substrate, on the solar cell reduced the efficiency of the solar cell by around 1%. Once we measure the power conversion efficiency with our cells on a glass substrate, we see that the efficiency is not different from the efficiency of the solar cell covered with solely the glass substrate. The efficiency has dropped about 1% compared to the uncovered solar cell, but this is most likely due to the fact that we placed the cells on a glass substrate. From this we can conclude that adding the cells directly onto the solar cells does not have a measurable effect on the efficiency of the solar cells.

So why do the microlenses described in literature cause a 20% increase in the efficiency and our cells do not? We think this is most likely due to our experimental setup. When conventional microlenses are placed onto a solar cell, the cells are not placed directly onto the solar cell, but there is a short distance (an ‘optical spacer’) between the microlens and the solar cell (Dal Zilio, Tvingstedt, Inganäs, & Tormen, 2009; Tvingstedt, Dal Zilio, Inganäs, & Tormen, 2008). This optical spacer is at the distance of the focal point of the microlens, allowing the microlens to focus the light on the cell. We however, have directly placed our microlenses onto a glass slide, which was immediately placed on the solar cell. Modeling of our biolenses has shown that our biolenses have a focal point at around 1 µm behind the cell. This means that if we are able to implement an optical spacer of 1 µm, we could focus the light on the surface of the solar cells. We now put the microlenses directly on the surface of the glass slide, which has its own optical properties, therefore we did not focus the light on the solar cell, as shown in figure 3.

Unfortunately we are not able yet to place our lenses in a structured array that is rigid enough to place it onto a solar cell with an optical space in between. This would be an essential future study if we want to further develop a technique where we want to improve solar cells. This is also applicable to our microlenses in optics and microscopy. For now, we have shown that our microlenses do not have any detrimental effects on the efficiency of solar cells. Most likely, this will also be the case for applications of our microlenses in optics. However, when we want to research improving solar cells or microscopy using our biolenses, we should first invest time into developing a technique to make a structured microlens array that is able to be placed onto a solar cell with an optical spacer in between.

Back to Top